[gambit-list] Any plans to support alternative IO mechanisms? (epoll, kqueue etc)
Eric Merritt
cyberlync at gmail.com
Wed Apr 27 13:59:29 EDT 2005
Marc,
I have neither convincing arguments nor great need. Its just
curiosity on my part and the inclination to provide answers for a
friend. Thanks for taking the time to formulate a response.
On 4/27/05, Marc Feeley <feeley at iro.umontreal.ca> wrote:
> There are no plans to use epoll/kqueue. However, the I/O runtime
> system has been written in a style that allows easily changing
> the underlying I/O synchronization mechanism. Currently "select"
> is used on POSIX systems, and MsgWaitForMultipleObjects on
> Windows, but it would be fairly easy (a day or two at most)
> to include support for another mechanism.
>
> The function ___device_select in os_io.c will need rewriting,
> and also ___device_select_add_fd.
>
> I'm driven by need, and currently select is good enough. So
> if you feel strongly about epoll/kqueue you have to bring some
> convincing arguments and show a real need.
>
> Marc
>
>
> On Apr 26, 2005, at 6:28 PM, Eric Merritt wrote:
>
> > A friend posed this question to me and I thought I would forward it
> > to the list to get a correct answer. Are there any plans to make use
> > of mechanisms like epoll or kqueue where they are available (on their
> > respective platforms)?
> > _______________________________________________
> > Gambit-list mailing list
> > Gambit-list at iro.umontreal.ca
> > http://mailman.iro.umontreal.ca/mailman/listinfo/gambit-list
>
>
More information about the Gambit-list
mailing list